Using deep learning techniques to measure neutrino oscillations

Important Note: Work from this project is migrating over to (title pending...) which will be the most up-to-date version. This repo will not be modfied further. (08/07/2024)

---

Project Supervisor: Prof. Ryan Nichol

Background

The NuMI (Neutrino at Main Injects) Off-axis $\nu_\text{e}$ Appearance (NOνA) experiment is a long baseline accelerator neutrino experiment that observes $\nu_\mu$ (or $\bar{\nu}_\mu$) disappearance and $\nu_\text{e}$ (or $\bar{\nu}_\text{e}$ ) appearance. NOνA consists of the 290 ton near detector and the 14 kton far detector that are located 1 km and 810 km away from NuMI respectively.

Figure 1: The NOvA far detector. (Credit: Fermilab)

Figure 2: Live event display of the NOvA far detector. Updates when you refresh the page. (Credit: Fermilab)

More information here.

Project aim

Improve the selection of $\nu_\mu$ events at the far detector using deep learning techniques. The current analysis procedure applies strict and inefficient cuts based on track containment (also, cuts based on CVN PID score).